University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln
USDA National Wildlife Research Center - Staff
Publications
Wildlife Damage Management, Internet Center for
2014
Winter Space Use of Coyotes in High-Elevation
Environments: Behavioral Adaptations to DeepSnow Landscapes
Jennifer L.B. Dowd
Eric M. Gese
Lise M. Aubry
Follow this and additional works at: http://digitalcommons.unl.edu/icwdm_usdanwrc
Part of the Life Sciences Commons
Dowd, Jennifer L.B.; Gese, Eric M.; and Aubry, Lise M., "Winter Space Use of Coyotes in High-Elevation Environments: Behavioral
Adaptations to Deep-Snow Landscapes" (2014). USDA National Wildlife Research Center - Staff Publications. Paper 1396.
http://digitalcommons.unl.edu/icwdm_usdanwrc/1396
This Article is brought to you for free and open access by the Wildlife Damage Management, Internet Center for at DigitalCommons@University of
Nebraska - Lincoln. It has been accepted for inclusion in USDA National Wildlife Research Center - Staff Publications by an authorized administrator
of DigitalCommons@University of Nebraska - Lincoln.
J Ethol (2014) 32:29–41
DOI 10.1007/s10164-013-0390-0
ARTICLE
Winter space use of coyotes in high-elevation environments:
behavioral adaptations to deep-snow landscapes
Jennifer L. B. Dowd • Eric M. Gese
Lise M. Aubry
•
Received: 14 May 2013 / Accepted: 9 October 2013 / Published online: 27 October 2013
Ó Japan Ethological Society and Springer Japan (outside the USA) 2013
Abstract In the last century, coyotes (Canis latrans)
have expanded their range geographically, but have also
expanded their use of habitats within currently occupied
regions. Because coyotes are not morphologically adapted
for travel in deep snow, we studied coyote space use patterns in a deep-snow landscape to examine behavioral
adaptations enabling them to use high elevations during
winter. We examined the influence of snow depth, snow
penetrability, canopy cover, and habitat type, as well as the
rates of prey and predator track encounters, on coyote
travel distance in high-elevation terrain in northwestern
Wyoming, USA. We backtracked 13 radio-collared coyotes
for 265.41 km during the winters of 2006–2007 and
2007–2008, and compared habitat use and movement patterns of the actual coyotes with 259.11 km of random
travel paths. Coyotes used specific habitats differently than
were available on the landscape. Open woodlands were
used for the majority of coyote travel distance, followed by
mixed conifer, and closed-stand spruce–fir. Prey track
encounters peaked in closed-stand, mature Douglas fir,
followed by 50- to 150-year-old lodgepole pine stands, and
0- to 40-year-old regeneration lodgepole pine stands.
Snowmobile trails had the most variation between use and
availability on the landscape (12.0 % use vs. 0.6 % available). Coyotes increased use of habitats with dense canopy
cover as snow penetration increased and rates of rodent and
J. L. B. Dowd L. M. Aubry
Department of Wildland Resources, Utah State University,
Logan, UT 84322-5230, USA
E. M. Gese (&)
U.S. Department of Agriculture, Wildlife Services,
National Wildlife Research Center, Utah State University,
Logan, UT 84322-5230, USA
e-mail: [email protected]
red squirrel track encounters increased. Additionally, coyotes spent more time in habitats containing more tracks of
ungulates. Conversely, use of habitats with less canopy
cover decreased as snow depth increased, and coyotes
traveled more directly in habitats with less canopy cover
and lower snow penetration, suggesting coyotes used these
habitats to travel. Coyotes persisted throughout the winter
and effectively used resources despite deep snow conditions in a high-elevation environment.
Keywords Canis latrans Coyote Habitat Snow
compaction Snow penetrability Space use
Introduction
Carnivore persistence in deep snow habitats is reliant on
their ability to maximize energetic trade-offs (Poulle et al.
1995; Crete and Lariviere 2003; Zub et al. 2009). Resource
selection is dependent on balancing energy expenditures
associated with locomotion versus energy intake from prey
while minimizing predation risk. Deep snow and cold
temperatures, both characteristic of harsh winter climates,
can exacerbate locomotion costs for cursorial predators
(Shield 1972; Crete and Lariviere 2003) causing a high
energetic budget and the need for acquiring substantial
food resources. Because of these energetic demands,
behavioral and/or morphological adaptations are necessary
for a species to effectively travel, hunt, and exploit
resources within such deep snow habitats, as demonstrated
in Canada lynx (Lynx canadensis) and snowshoe hare
(Lepus americanus; Murray and Boutin 1991; Lesage et al.
2001; Murray and Larivière 2002).
Coyote (Canis latrans) encroachment into deep-snow
landscapes is a concern because of their association with
123
30
snowmobile compacted trails, and subsequent possible
competition with lynx (Murray and Boutin 1991; Koehler
and Aubry 1994; Murray et al. 1995; Lewis and Wenger
1998; Bunnell et al. 2006). Although one study found snow
compaction did not result in competition between coyotes
and lynx (Kolbe et al. 2007), other studies suggest that
geographically distinct regions differing in snow profile,
predator communities, and expanse of snow compaction
resulting from snowmobile use could result in different
findings (Bunnell et al. 2006). Canids may demonstrate a
higher level of energetic tolerance in response to deep
snow (Crete and Lariviere 2003). Behavioral traits facilitating coyote use of deep-snow habitats include their ability
to actively select travel paths with shallower, more supportive snow (Murray and Boutin 1991; Kolbe et al. 2007);
flexibility in prey selection and feeding habits (Patterson
et al. 1998; Bartel and Knowlton 2004); and hunting in
groups to acquire larger prey (Gese and Grothe 1995).
Studies have observed behaviors of coyotes dwelling in
deep-snow habitats (Murray and Boutin 1991; Litvaitis
1992; Crete and Lariviere 2003; Thibault and Ouellet
2005). However, few have examined how coyotes use the
landscape from a spatial perspective, and how extrinsic
factors such as snow depth, snow supportiveness, prey
availability, canopy cover, and habitat type influences
space use. Although a recent study investigated the influence of groomed trails on coyote movements (Kolbe et al.
2007), no studies to date have specifically analyzed the
influence of groomed trails on habitat use within specific
cover types.
Our objective was to document space use by coyotes in
high-elevation terrain characterized by long winters and
deep snow to determine what variables influenced coyote
use of deep-snow environments, and to understand what
enables year-round persistence under presumably unfavorable conditions. Accordingly, we examined variables
encountered within specific habitats and compared coyote
use of those habitats to availability across the landscape.
Specifically, we were interested in understanding how
snow characteristics (snow depth and supportiveness),
canopy cover, habitat type, prey track encounter rates, and
predator track encounter rates influenced coyote travel
distance in different habitats. We hypothesize that (1)
coyotes will select for groomed trails to travel to and from
sites that are rich in prey; however, (2) the benefits of
increased prey encounters within habitats with high snow
penetration could outweigh the costs of travelling on
unsupported snow. Here, we will address these competing
hypotheses by establishing first whether coyotes preferentially use groomed trails to travel by comparing availability versus use of groomed trails, and by quantifying
which habitats are preferentially selected based on characteristics such as snow characteristics (snow depth and
123
J Ethol (2014) 32:29–41
supportiveness), canopy cover, habitat type, prey track
encounter rates, and predator track encounter rates. We
predict that (1) coyotes will increase their use of snowmobile trails as snow depth increases and snow penetration increases (i.e., snow compaction decreases), (2)
coyotes will increase use of habitats with high canopy
cover as this canopy cover suspends the snow in the
canopy and reduces snow cover under the canopy, (3)
coyotes will increase their use of habitats containing high
prey abundance, (4) coyotes will decrease their use of
habitats containing larger sympatric predators (i.e.,
wolves, Canis lupus), and (5) the pattern of movement
(convoluted versus straight-line travel) of coyotes will be
influenced by habitat type, canopy cover, snow depth and
penetrability, and prey and predator encounter rates, with
the hypothesis that coyotes will travel in a more convoluted path while foraging, versus moving in a straight path
when traveling between resource patches.
Materials and methods
Study area
We conducted this study on the east and west sides of
Togwotee Pass in northwestern Wyoming, USA. The
512-km2 study area was characterized by extensive recreational trails and roads (*2 km/km2) maintained yearround. The area was composed of the Bridger-Teton and
Shoshone National Forests, and privately owned ranches.
Elevations ranged from 1,800 to [3,600 m. The area was
characterized by short, cool summers (mean temperature of
12 °C) and long winters (mean temperature of -8 °C).
Precipitation occurred mostly as snow, and mean maximum snow depths ranged from 100 cm at lower elevations
to [245 cm at intermediate elevations (2,000–2,400 m).
Cumulative monthly snow depth for the winter (December–April) averaged 226.6, 149.4, and 228.9 cm in 2006,
2007, and 2008, respectively (Natural Resources Conservation Service 2008). Habitats varied between the east and
west sides of the pass, with the east side classified as dry
and the west side as wet (US Forest Service 1989). The
plant communities on the east side consisted of more open
dry meadows, sagebrush (Artemisia spp.), and stands of
lodgepole pine (Pinus contorta), while the west side had
more wet meadows and stands of Douglas fir (Pseudotsuga
menziesii) and spruce (Picea engelmannii). On both sides
of the pass, the plant communities included cottonwood
(Populus angustifolia) riparian zones, interspersed with
sagebrush uplands and willow (Salix spp.)—wetland
communities at lower elevations. At intermediate elevations, aspen (Populus tremuloides), Douglas fir, and
lodgepole pine were the dominant species. Whitebark pine
J Ethol (2014) 32:29–41
(Pinus albicaulis), spruce, and sub-alpine fir (Abies lasiocarpa) were the primary tree species at higher elevations.
The area around Togwotee Pass was a complex ecosystem with a diverse assemblage of predators. Although
wolves (C. lupus) were extirpated from Wyoming by the
1930s, they have since re-established due to the 1995 reintroduction into Yellowstone National Park (US Fish and
Wildlife Service 2006). Other carnivores included cougar
(Puma concolor), wolverine (Gulo gulo), grizzly bear
(Ursus arctos), black bear (U. americanus), bobcat (L.
rufus), red fox (Vulpes vulpes), and pine marten (Martes
americana). The main competitor and predator of coyotes
was the wolf, while coyotes competed with bobcats, lynx,
and red foxes for similar prey resources and habitat.
Ungulate species found in the area included elk (Cervus
elaphus), moose (Alces alces), bison (Bison bison), bighorn
sheep (Ovis canadensis), mule deer (Odocoileus hemionus), and white-tailed deer (O. virginianus). Pronghorn
antelope (Antilocapra americana) were in the area only
during the summer. Other species included snowshoe
hares, red squirrels (Tamiasciurus hudsonicus), Uinta
ground squirrels (Spermophilus armatus), black-tailed
jackrabbits (L. californicus), cottontail rabbits (Sylvilagus
spp.), ruffed grouse (Bonasa umbellus), blue grouse
(Dendragapus obscurus), deer mice (Peromyscus maniculatus), voles (Microtus spp.), gophers (Thomomys spp.),
and various cricetid species. The main prey items in the
diet of coyotes on the study area included mule deer
(20.1 % occurrence), elk (12.5 %), montane vole (Microtus
montanus, 12.0 %), and snowshoe hares (8.0 %) as found
via scat analysis (Dowd and Gese 2012).
Snowmobiling was extensive during winter, allowing
riders to access groomed trails and off-trail riding in and
around the study area once snow conditions permitted (late
October–May). Trail grooming operations typically began
by mid-December with trails maintained through April 1
depending on snowfall. Wyoming’s Continental Divide
Snowmobile Trail was considered one of the top trail
systems in the west (Wyoming Department of State Parks
and Cultural Services 2008).
Habitat classifications
For our study area, habitat types were categorized
according to vegetation age, stand structure, and species
composition based on direct observation by field personnel
during travel path sampling (see next section). Due to the
scale of our study and the inadequacy of GIS layers currently available for the area, we used a vegetation classification system that combined dominant tree species and
the stand’s successional stage, representing a distinct
‘cover type’ (Despain 1990). Much of the variation in stand
age was due to historic logging, fires, and other natural
31
disturbances (e.g., disease, avalanches, high winds). Cover
types used a two-letter code paired with a number to
classify a continuous patch (e.g., LP for lodgepole pine, 0
for a young stand = LP0). Lower numbers represented
younger stands while higher numbers represented older
stands; 0 = 0–40 years, 1 = 50–150 years, 2 = 150–300
years, and 3 = 300? years old. A two-letter abbreviation
lacking an attached number represented a climax stand
(i.e., final successional stage). Specific cover types in our
study area included aspen–conifer (AC), aspen (AS),
Douglas-fir (DF0–DF3), lodgepole pine (LP0–LP3), mixed
conifer (MC), open woodland (OW), spruce–fir (SF0–SF1),
and whitebark pine (WB0–WB2). For the purpose of this
study, we also classified groomed trail (GT) as a distinct
habitat classification. Using this system, we documented a
total of 20 distinct habitat types in our study area.
Coyote capture and backtracking
We captured coyotes in the summer and fall using paddedjaw leg-hold traps with attached tranquilizer tabs. We also
captured coyotes during winter by placing road-killed deer
and elk carcasses in large open meadows and using
snowmobiles with nets, or net-gunning from a helicopter
(Gese et al. 1987). Coyotes were radio-collared, ear-tagged, weighed, and released at the capture site; animals were
handled without immobilizing drugs. Research protocols
were approved by the Institutional Animal Care and Use
Committees at Utah State University (#1294) and the
USDA/National Wildlife Research Center (QA-1389).
We backtracked radio-collared coyotes during the winter months of 2006–2007 and 2007–2008 following
methods of Kolbe et al. (2007) to document habitat use and
spatial patterns on snow-compacted routes and non-compacted terrain (i.e., areas not used by snowmobiles). We
used data collected during the backtracking to determine
how extrinsic factors (prey and predator track encounter
rates, snow depth, snow penetration, canopy cover, and
habitat type) influenced the distance a coyote traveled
within a given habitat. We randomly selected individual
coyotes for backtracking using a computer generated randomization sequence (SAS Institute, Cary, NC, USA) to
avoid bias and ensure all coyotes were sampled equally.
The night before a backtracking session, we located coyotes by triangulation using C3 azimuths, and their position
projected using LOCATE II, v.1.82 (Nova Scotia Agricultural College, Truro, Nova Scotia, Canada). Once the
travel path location was verified, a starting location for the
actual travel path was used to generate a starting point for
the random travel path. These random travel paths we
generated allowed for direct comparison to the actual
coyote travel paths and thus assess habitat selection. We
created random travel paths using digital layers from
123
32
previously documented coyote travel paths in a random
direction and projection (or ‘‘spin’’), 2–3 km distance from
the actual start point of the individual being backtracked
that day. We used a projection distance to ensure sampling
independence between the actual and random travel paths
(Kolbe et al. 2007).
We began backtracking in the morning after night
movements of coyotes had occurred and before the snow
column deteriorated. Both actual and random paths were
tracked simultaneously by teams of 2 field personnel,
taking measurements and recording data for C3 km of
tracking. During each actual or random travel path, we
used a hand-held computer (Trimble GeoExplorer, Sunnyvale, California, USA) to collect data in a digital format
using a datasheet generated with the computer software
GPS Pathfinder Office (Trimble Navigation, Westminster,
CO, USA). During each actual or random travel path,
pathfinder software recorded locations every 5 s along the
travel path. We marked point locations every time a habitat
change was encountered, organizing the travel path into
distinct but consecutive segments identified by habitat
(Kolbe et al. 2007). We recorded canopy cover within each
habitat using a densiometer to rank canopy closure into 4
categories: 0–10, 11–39, 40–69, and 70–100 % canopy
cover. We recorded prey and predator track crossings as
point locations by number and species every time a set of
animal tracks crossed a travel path. We measured snow
depth at every habitat change and every 200 m along the
travel path using a probe (marked in cm) to measure from
the snow surface to the ground. We recorded snow penetration whenever the habitat changed and every 200 m
along the travel path by dropping a 100-g weight from 1 m
above the snow surface and measuring penetration (Kolbe
et al. 2007). Once the travel paths were completed, data
recorded on the Trimble units were downloaded and
imported into GPS Pathfinder Office, then differentially
corrected. Travel paths were then smoothed to eliminate
bounce or GPS scatter caused by canopy cover or varying
topography which can influence location accuracy. We
converted all travel paths to ArcGIS files for analysis.
Data and statistical analyses
We measured coyote habitat use at the landscape level by
classifying the relative proportion of 20 habitats randomly
encountered throughout the study area and comparing the
habitats used by coyotes on actual travel paths (Thibault
and Ouellet 2005). Randomly encountered habitats were
documented along random travel paths in the same manner
that habitats were encountered and recorded along simultaneously conducted actual travel paths of a coyote. Distances were referred to as the ‘control’ (random distance)
and the ‘treatment’ (actual distance). Due to unequal
123
J Ethol (2014) 32:29–41
sample sizes resulting from differences in habitat encounters between actual and random travel paths, we used Levene’s test to assess the equality of variance in and between
habitats. Unequal variances led to the use of a nonparametric Kruskal–Wallis test with Bonferroni corrections of
P values in R v.2.10.1 (R Development Core Team 2010;
package ‘Agricolae’, ‘Kruskal’ procedure) to compare
differences across habitat types between the control and
treatment groups, as well as differences within habitat
types (actual distance traveled by coyotes within each
habitat type to distance within control sites). All comparisons with a P value B0.10 were considered significant. All
distance means and standard errors (SE) were presented for
habitat types within control and treatment groups.
The covariates we hypothesized to be most important in
determining how coyotes used the landscape included
habitat characteristics (habitat ‘HAB’, canopy cover ‘CC’),
snow characteristics (snow depth ‘SD’, snow penetration
‘SP’), predator track encounters (wolf ‘WF’), and prey
track encounters (snowshoes hares ‘SSH’, red squirrels
‘RS’, grouse ‘GR’, rodents ‘ROD’, ungulates ‘UNG’). We
separated red squirrels from rodents as the squirrels were
principally arboreal prey, while most of the rodents were
microtines. Ungulates were grouped to improve sample
size (e.g., mule deer, elk, and moose). As an alternative to
considering all of the prey species additively, we considered another covariate accounting for total prey abundance
‘TotPrey’, in an attempt to conserve degrees of freedom in
the analysis conducted below. Because some of the
covariates had the potential to be collinear (i.e., strongly
correlated), we calculated variance inflation factor (i.e.,
package ‘car,’ procedure ‘vip’ in R v.2.10.1; R Development Core Team 2010) across covariates prior to model
selection (Neter et al. 1996). A variance inflation factor
of \5 indicated a lack of colinearity, and vice versa. We
conducted all the analyses below in R v.2.10.1 (R Development Core Team 2010).
Distance traveled within various habitats allowed us to
examine movement patterns (convoluted use versus straight
line use) and understand the behaviors associated with how
coyotes used these habitats. To understand which of the
above-mentioned factors could explain variability in the
distance covered by coyotes within a given habitat, we
compared actual distance traveled within a habitat segment
to the shortest possible distance between the entrance and
the exit points of that habitat. A distance ratio was then
calculated by dividing the shortest possible distance by the
actual distance traveled by a coyote, providing us with a
proportion that ranged from 0 to 1, (i.e., ‘LRATIO’ =
shortest distance/actual distance). This measure might seem
counter intuitive since we would usually be interested in the
distance covered by a coyote relative to the shortest possible
distance; however, we needed this ratio to be constrained
J Ethol (2014) 32:29–41
between 0 and 1 in order to be able to conduct betaregressions. This ratio is a reasonable proxy to the time
spend in a given area, and thus can help us learn more about
foraging behaviors across habitat types and snow characteristics, and as a function of both predator and prey
encounter rates. To address this, we used a modification of
the ‘‘empirical logistic transform’’ proposed by Collett
(2002) for data that are not discrete. Such modification of
the logistic regression is recommended in situations where
the dependent variable (LRATIO) is continuous and
restricted to the unit interval 0–1, such as proportions or
rates. The dependent variable needs to be logit-transformed,
such as log[LRATIO/(1 - RATIO)], prior to conducting
linear regression with an identity link (Warton and Hui
2011). We also included an individual random effect to
account for both repeated measures across individuals, and
spatial auto-correlation. Even though tracks were measured
repeatedly in space within the same home range for a given
individual, by accounting for an individual random effect
that controls for repeated track measurements within an
individual’s home range, we solve both the issue of individual and spatial autocorrelation at once. We used generalized linear mixed models (GLMMs) (Kuznetsova et al.
2013; package ‘lmerTest’, procedure ‘lmer’) to model the
effects of various covariates on LRATIO, while accounting
for both individual and spatial auto-correlation.
Due to the influence of canopy cover on snowpack
accumulation in forests (Bernier and Swanson 1992; Murray and Buttle 2003; Talbot et al. 2006), we assessed coyote
habitat use by comparing variables (snow characteristics,
prey and predator encounters) documented along actual
coyote travel paths using another habitat structure variable,
canopy cover, within 4 levels: 0–10, 11–39, 40–69, and
70–100 % (percent’s reflect increased canopy closure). To
determine differences in canopy cover use by coyotes, we
analyzed the use of various CC measures, as a function of
snow characteristics (i.e., snow depth ‘SD’, snow penetration ‘SP’), predator track encounters (i.e., wolf ‘WF’), and
prey track encounters (i.e., snowshoes hares ‘SSH’, red
squirrels ‘RS’, grouse ‘GR’, rodents ‘ROD’, ungulates
‘UNG’). As an alternative to considering all of the prey
species additively, we again considered another covariate
accounting for total prey encountered ‘TotPrey’. All
explanatory covariates were treated as continuous, and the
response variable, CC, was treated as an ordinal categorical
variable (CC = 1 if canopy cover was between 0 and 10 %,
CC = 2 if between 11 and 39 %, CC = 3 if between 40 and
69 %, and CC = 4 if [69 %). Here, we also used GLMMs
(Kuznetsova et al. 2013; package ‘lmerTest’, procedure
‘lmer’) to model the effects of various covariates on a
coyote’s choice of canopy cover levels (CC) while
accounting for individual auto-correlation.
33
Model selection
For both sets of analyses described above, we defined a
global model testing for additive and interactive effects of
all of the covariates of interest, but only when they made
biological sense. For analysis where ‘LRATIO’ was the
response variable, we specifically considered interactions
between habitat and snow characteristics (i.e., canopy
cover ‘CC’, snow depth ‘SD’, snow penetration ‘SP’) and
predator or prey track encounters (e.g., wolf ‘WF’, snowshoes hares ‘SSH’, red squirrels ‘RS’, grouse ‘GR’, rodents
‘ROD’, ungulates ‘UNG’). For analyses where ‘CC’ was
the response variable, we considered interactions between
snow characteristics and predator or prey track encounters.
Since the presence of predators and availability of prey
could strongly depend on snow conditions, interactions
between these sets of variables could help explain coyote
habitat use.
We used a unique approach to model selection based on
parameter significance (confidence intervals and P values)
alone. Since we are interested in objectively testing
hypotheses about which factors are important in explaining
coyote habitat use, we simply fit a full model accounting
for all biological variables of interest, as well as biologically relevant interaction terms, and base our inference
solely on parameter estimates, associated confidence
intervals and P values from the full model (Bolker 2008).
For each estimated parameter (bi) that appeared in the
model, we assessed the precision of each bi based on the
extent to which 95 % confidence intervals for each bi
overlapped zero (Graybill and Iyer 1994), and associated
P values, to discuss the significance of each covariate
effect on the response variable (either LRATIO, or CC).
Results
A total of 15 (4 F, 11 M) coyotes were captured and radiocollared from August 2006 through February 2008. One
individual was shot shortly after being radio-collared and 1
young coyote dispersed from the study area, leaving 13
individuals (4 F, 9 M) for sampling. A total of 59 coyote
travel paths were followed for a combined distance of
265.43 km, for 1,154 individual habitat segments. We also
collected 259.11 km of random travel paths (1,426 individual habitat segments) for comparative analysis. Although
20 distinct cover types were documented throughout the
study area, only 18 habitats were encountered by coyotes
(DF and DF1 were not used by coyotes). Additionally, one
habitat type was encountered by coyotes, but not encountered on our control (random) surveys (WB1).
123
34
J Ethol (2014) 32:29–41
Coyote habitat selection and use
We compiled a ranking system based on how habitats were
used. Assumptions regarding what criteria make a habitat
desirable to a coyote were made to rank each habitat by the
number of prey encounters, predator encounters, snow
depth, snow penetration, and travel distance ratio
(Table 1). Coyotes used open woodlands (OW) for the
Table 1 Habitat use by coyotes (Canis latrans) from actual backtrack data in the Togwotee Pass study area, northwestern Wyoming, winters
2006–2007 and 2007–2008
Habitat type
Habitat
use (%)
Habitat
availability
(%)
Prey
encounters
(#/km)
Predator
encounters
(no./km)
Snow depth
(shallowest to
deepest)
Snow penetration (most
supportive to least
supportive)
Ratio (start at lowest
ratio = most to least
hunting)
OW (Open
woodland)
(1) 25.6
(1) 38.2
(16) 10.6
(18) 1.3
(16) 97.6
(4) 16.9
(16) 0.61
MC (Mixed
conifer)
(2) 21.3
(2) 20.8
(6) 34.5
(13) 0.3
(12) 90.6
(13) 21.2
(6) 0.43
GT (Groomed
trail)
(3) 12.0
(13) 0.6
(8) 29.0
(1) 0.0
(1) 42.2
(1) 4.9
(18) 0.70
SF1 (Spruce–fir
closed stand)
(4) 9.6
(4) 8.5
(5) 36.8
(1) 0.0
(10) 88.6
(17) 24.0
(2) 0.36
LP2 (Lodgepole
Pine
150–300 years)
(5) 6.5
(6) 3.4
(13) 22.5
(14) 0.6
(7) 81.6
(6) 17.8
(8) 0.44
SF0 (Spruce–fir
open canopy)
(6) 4.9
(3) 9.1
(11) 23.6
(15) 1.0
(14) 91.2
(11) 20.4
(8) 0.44
LP1 (Lodgepole
pine
50–150 years)
(7) 4.6
(5) 5.8
(2) 46.8
(1) 0.0
(6) 81.0
(8) 19.9
(13) 0.52
LP3 (Lodgepole
pine 300?
w/spruce)
(8) 4.0
(7) 2.6
(10) 24.6
(16) 1.0
(9) 82.4
(15) 22.7
(4) 0.41
SF (Spruce–fir
climax
w/WBP)
(9) 3.9
(8) 2.6
(14) 18.2
(1) 0.0
(5) 80.1
(16) 23.1
(3) 0.37
AC (Aspen/
conifer)
(10) 3.7
(9) 1.9
(12) 22.7
(17) 1.0
(11) 89.0
(10) 20.1
(12) 0.50
LP0 (Lodgepole
pine
0–40 years)
(11) 1.6
(10) 1.9
(3) 41.1
(1) 0.0
(17) 104.0
(7) 19.5
(15) 0.54
WB3 (Pole to
mature)
(12) 0.7
(16) 0.4
(9) 26.1
(1) 0.0
(2) 49.0
(12) 21.1
(1) 0.29
WB2 (Mature,
codominance)
(12) 0.7
(14) 0.6
(15) 15.9
(1) 0.0
(13) 91.2
(18) 24.7
(7) 0.43
AS (Aspen, all
age)
(14) 0.5
(11) 1.5
(7) 29.5
(1) 0.0
(8) 81.7
(3) 16.6
(14) 0.53
WB1 (Whitebark
pine, pole)
(15) 0.2
(18) 0.0
(17) 5.9
(1) 0.0
(18) 110.0
(14) 22.0
(10) 0.49
LP (Lodgepole
Pine 300?
climax)
(16) 0.1
(15) 0.6
(4) 37.9
(1) 0.0
(15) 97.2
(5) 17.0
(17) 0.63
DF2 (Douglas
fir, closed,
mature)
(17) 0.1
(12) 0.9
(1) 108.1
(1) 0.0
(4) 75.5
(9) 20.0
(5) 0.42
WB (All
whitebark,
overmature)
(18) 0.0
(17) 0.2
(18) 0.0
(1) 0.0
(3) 59.0
(2) 8.0
(10) 0.49
Ranking system based on assumption from the most desirable habitats (1 = most desirable, 18 = least desirable) reflecting observed encounters
on actual travel paths shown in parentheses
123
J Ethol (2014) 32:29–41
majority of their travel distance, followed by mixed conifer
(MC), and closed-stand spruce–fir (SF1). Prey encounters
peaked in closed canopy, mature Douglas fir (DF2), followed by dense, young lodge pine (LP1), recently burned
lodgepole pine (LP0), climax stand lodgepole pine (LP),
and closed-stand spruce–fir (SF1). Wolf track crossings
were most frequent in open woodland (OW), aspen conifer
and 300? year successional forest lodgepole pine (LP3).
There were no wolf tracks crossed by coyotes in the
majority of all habitat types (GT, SF1, LP1, SF, LP0, WB3,
WB2, AS, WB1, LP or WB). The habitats with the shallowest snow were GT, mature whitebark pine co-dominated with spruce, fir, and lodgepole (WB3) and mature
whitebark pine (WB). The most supportive snow was also
on GT, followed by mature and AS. The greatest travel
distance ratio was encountered in GT, meaning coyotes
spent the least amount of time deviating from their projected entrance to exit points in this habitat. A high ratio
demonstrated a straighter travel path, compared to a lower
ratio which demonstrates convolutions in the travel path.
Climax stands of lodgepole pine (LP) and open woodlands
(OW) had the next highest distance ratios (Table 1).
When comparing habitats encountered on control paths
to actual coyote backtracks, there were three habitat types
that were not encountered in either dataset: WB1, DF, and
DF1. For comparative purposes, these habitats were
removed from the analysis. For the most part, coyote use of
habitats ranked similarly to availability. The most readily
available habitat across our study area based on our random
travel paths was open woodland, followed by mixed
conifer, young spruce–fir, and closed-stand spruce–fir
(Table 1). Almost all the top ten ranking habitats used by
coyotes were also in the top ten habitats available across
the landscape (Table 1). However, according to our random path analysis, groomed trail was only available 0.6 %
of the time, while it ranked third in coyote use, accounting
for 12.0 % of their travel distance, meaning that, proportionally, they used groomed trails 18.5 times more than
available. This was notably higher than any other habitat
type encountered on the landscape. We further confirmed
this result by conducting a Kruskal–Wallis test which
indicated a significant difference in distance covered
between the control (random distance) and the treatment
(actual distance), across habitats (X2 = 154.39, df = 16,
P \ 0.001) and between habitats (Table 2). Coyotes used
only GT, LP2, and WB3 significantly more than expected,
while they used only LP1 and SF0 less than expected
(Table 2).
Effect of snow, prey, and predators on distance traveled
We did not experience any issues with colinearity when
conducting regression analyses; all variance inflation
35
Table 2 Kruskal–Wallis test for differences in habitat use between
the actual distance covered by the coyotes and the random distance,
across 17 habitat types in the Togwotee Pass study area, northwestern
Wyoming, winters 2006–2007 and 2007–2008; three habitats were
excluded due to lack of encounter on either actual or random travel
paths
Habitat
Random
Actual
Mean
SE
AC
0.160
0.109
AS
DF2
0.127
0.171
0.084
0.072
n
P
Mean
SE
n
31
0.286
0.354
34
0.215
30
13
0.181
0.121
0.139
0.025
38
2
0.023
0.248
GT
0.038
0.042
47
0.339
0.450
94
\0.001
LP
0.317
0.220
5
0.094
0.110
3
0.101
LP0
0.164
0.118
30
0.166
0.129
25
0.906
LP1
0.173
0.120
87
0.128
0.110
95
0.005
LP2
0.237
0.160
37
0.223
0.261
77
0.041
LP3
0.254
0.198
27
0.325
0.448
33
0.247
MC
0.250
0.231
216
0.254
0.266
222
0.246
OW
0.149
0.097
666
0.202
0.268
337
0.457
SF
0.416
0.397
16
0.338
0.563
31
0.181
SF0
0.239
0.170
99
0.196
0.197
67
0.020
SF1
0.230
0.206
96
0.256
0.247
100
0.881
WB
0.123
0.062
5
0.033
–
1
0.441
WB2
0.232
0.250
7
0.164
0.110
11
0.598
WB3
0.093
0.065
12
0.245
0.236
7
0.305
factors were \5 (Table 3; Neter et al. 1996). Results pertaining to the generalized linear mixed model testing for
the effect of all covariates of interest on distance ratio
‘LRATIO’ while controlling for individual and spatial
autocorrelation are presented (Table 4). The following
covariates and interactions had a significant relationship
with LRATIO: rodent track encounters ‘ROD’ and an
interaction between snow depth and grouse track encounter
‘SD 9 GR’ both had a positive relationship with distance
ratio (Table 4). However, CC, SP, an interaction between
snow depth and rodent track encounters ‘SD 9 ROD’, as
well as an interaction between canopy cover and rodent
track encounter ‘CC 9 ROD’ all had a negative relationship with distance ratio (Table 4). These results indicated
coyotes covered less distance (compared to the shortest
possible distance and thus spent more time) in habitats with
dense canopy cover (Table 4) and similarly when snow
penetration increased, suggesting coyotes would tend to
cover more distance (i.e., spend less time) in locales where
snow penetration is low compared to the shortest possible
distance (Table 4), and that coyotes covered less distance
(i.e., spent more time) in ‘closed’ habitats (thick cover),
and more distance (i.e., less time) in open habitats (Fig. 1).
Coyotes tended to cover less distance (i.e., spend more
time) in areas with more rodents than needed, especially as
canopy cover and snow depth increased (Table 4).
123
36
J Ethol (2014) 32:29–41
Table 3 Variance inflation factors presented for all variables considered in A beta regressions testing for the effects of habitat, snow,
and prey characteristics on LRatio (i.e., shortest possible distance by
the actual distance traveled by a coyote), and B generalized linear
mixed models testing for the effects of habitat, snow, and prey
characteristics on CC (canopy cover categories)
Explanatory variables
VIF
Table 4 Results pertaining to the best performing generalized linear
mixed model testing for the effects of habitat (canopy cover ‘CC’),
snow characteristics (snow depth ‘SD’, snow penetration ‘SP’), small
prey track crossing rates (snowshoe hare ‘SSH’, red squirrel ‘RS’,
grouse ‘GR’, rodent ‘ROD’), ungulate track crossing rates (‘UNG’),
and predator track crossing rates (wolf ‘WF) on the distance covered
by coyotes ‘LRATIO’
b Estimate
SE
t value
P value
A beta regressions
2.07
CCa
-0.3611
0.1733
-2.0840
0.0379
Canopy cover ‘CC’
1.15
Snow depth ‘SD’
1.13
SD
SPa
0.0010
-0.0210
0.0019
0.0099
0.5630
-2.1230
0.5737
0.0344
Snow penetration ‘SP’
1.13
SSH
0.0167
0.0106
1.5690
0.1175
Snow shoe hare ‘SSH’
Red squirrel ‘RS’
1.31
1.48
RS
0.0076
0.0349
0.2190
0.8268
GR
-0.3202
0.2333
-1.3730
0.1707
Grouse ‘GR’
1.03
RODa
0.0797
0.0473
1.6850
0.0930
Rodent ‘ROD’
1.17
UNG
-0.0492
0.0326
-1.5120
0.1315
Wolf ‘WF’
1.04
WF
-0.0543
0.1313
-0.4130
0.6797
0.6238
Habitat type ‘HAB’
CC 9 SSH
-0.0027
0.0055
-0.4910
1.45
SD 9 SSH
0.0001
0.0002
0.3300
0.7418
Snow depth ‘SD’
1.13
SP 9 SSH
-0.0007
0.0006
-1.1600
0.2468
Snow penetration ‘SP’
1.04
CC 9 RS
0.0016
0.0079
0.1970
0.8437
1.31
SD 9 RS
0.0000
0.0004
-0.0370
0.9707
Red squirrel ‘RS’
1.46
SP 9 RS
-0.0007
0.0009
-0.7370
0.4615
Grouse ‘GR’
1.03
CC 9 GR
0.0378
0.1484
0.2550
0.7992
Rodent ‘RD’
1.17
Wolf ‘WF’
1.04
SD 9 GRa
SP 9 GR
0.0028
0.0000
0.0016
0.0043
1.8280
0.0000
0.0684
1.0000
CC 9 RODa
-0.1129
0.0554
-2.0380
0.0423
SD 9 RODa
-0.0017
0.0009
-1.9000
0.0582
B generalized linear mixed models
Habitat type ‘HAB’
Snow shoe hare ‘SSH’
Habitat use within 4 levels of canopy cover
The results pertaining to the generalized linear mixed
model testing for the effect of all covariates of interest on
canopy cover ‘CC’ while controlling for individual and
spatial autocorrelation are presented (Table 5). The best
performing model indicated SD had a significant but weak
negative effect on canopy cover (Table 5), whereby the
deeper the snow, the larger the preference for low canopy
cover habitats (Fig. 2). Snow penetration had the opposite
effect on habitat use (Table 5); and as snow penetration
increased, the use of habitats where canopy cover was
dense increased as well (Fig. 2). The presence of wolves,
snowshoe hares, red squirrels, and ungulates all had a
significant effect on canopy cover as well, whereby the
higher the encounter rate of wolves and snowshoe hares,
the larger the preference for dense canopy covers by coyotes (Table 5; Fig. 2). We also considered biologically
meaningful interactions between covariates that revealed
interesting results: coyotes did not select for high canopy
cover when snow penetration was high, even in the presence of increased track encounters of both rodents and
ungulates (Table 5). Similarly, they avoided high canopy
cover when snow penetration and wolf track encounters
increased (Table 5). Finally, they selected for increased
123
SP 9 ROD
0.0093
0.0064
1.4520
0.1475
CC 9 UNG
-0.0039
0.0118
-0.3270
0.7436
SD 9 UNG
0.0003
0.0004
0.9420
0.3468
SP 9 UNG
0.0012
0.0013
0.9590
0.3383
CC 9 WF
0.0343
0.0572
0.5990
0.5493
SD 9 WF
0.0004
0.0014
0.2960
0.7677
SP 9 WF
0.0008
0.0019
0.4330
0.6650
a
Covariates that had a significant effect on LRATIO
canopy cover when snow depth and ungulate presence
increased (Table 5); note that the relationship between
canopy cover and snow depth mirrors that of canopy cover
and snow penetration (Fig. 2).
Discussion
Coyote habitat use versus availability
Although habitat rankings were similar in regards to what
was most and least used between random and actual habitat
encounters, our distance comparisons showed proportional
habitat use by coyotes did not reflect availability in the
landscape; in many cases, coyotes used specific habitats
J Ethol (2014) 32:29–41
37
Fig. 1 Relationship between
distance ratio (LRatio) and
significant biological covariates
of interest (Table 4):
relationships between distance
ratio and canopy cover ‘CC’
(top left panel), snow
penetration ‘SP’ (top right
panel), grouse ‘GR’ (bottom left
panel), and rodent encounter
rates ‘ROD’ (bottom right
panel)
Table 5 Results pertaining to the best performing generalized linear
mixed model testing for the effects of snow characteristics (snow
depth ‘SD’, snow penetration ‘SP’), prey track crossing rates (red
squirrel ‘RS’, grouse ‘GR’, rodent ‘ROD’, ungulates ‘UNG’), and
predator track (wolf ‘WF’) crossing rates on canopy cover ‘CC’
selection by coyotes, while accounting for both individual and spatial
auto-correlation
b Estimate
SE
t value
P value
SDa
-0.0020
0.0005
-3.7920
SPa
SSH
0.0189
0.0016
0.0020
0.0016
9.6270
0.9720
\0.001
0.3315
RS
0.0040
0.0037
1.0930
0.2746
GR
0.0169
0.0272
0.6210
0.5348
ROD
-0.0008
0.0039
-0.2150
0.8297
UNG
-0.0004
0.0026
-0.1500
0.8807
0.0508
0.0281
1.8080
0.0709
WFa
\0.001
SD 9 SSHa
0.0001
0.0000
-2.1040
0.0356
SP 9 SSH
0.0001
0.0001
1.2930
0.1963
SD 9 RS
0.0001
0.0000
1.3650
0.1726
SP 9 RSa
-0.0002
0.0001
-1.6620
0.0968
SD 9 GR
0.0000
0.0002
0.1940
0.8461
SP 9 GR
-0.0008
0.0009
-0.8520
0.3942
SD 9 ROD
-0.0001
0.0001
-0.9620
0.3363
SP 9 ROD
0.0001
0.0003
0.3190
0.7498
SD 9 UNGa
SP 9 UNGa
0.0002
-0.0004
0.0001
0.0001
2.8570
-3.0150
0.0044
0.0026
SD 9 WFa
-0.0005
0.0003
-1.7690
0.0771
SP 9 WF
-0.0006
0.0004
-1.4260
0.1542
a
Variables which have a significant effect on canopy cover selection
by coyotes
more or less than were available. Differences in distance
spent both between habitats and within habitats indicated
that landscape use was not random, but rather an active
selection process. Significantly more use of groomed trails,
lodgepole pine, and mature whitebark pine co-dominated
with spruce, fir, and lodgepole pine suggested that these
habitats provided desirable habitat features and associated
resources for coyotes.
Coyotes used groomed trails for a high proportion of
their travel distance compared to availability (12.0 vs.
0.6 %) suggesting coyotes may be selecting groomed trails
which could represent an important behavioral adaptation.
Based on our rankings of desirable habitats which considered individual variables and basic assumptions from
observed encounters (Table 1), we suspect the reason for
the high use of groomed trails compared to availability
could due to a low predator encounter rate, low snow
depth, and low snow penetration (of which GT received ‘1’
rankings for all the aforementioned variables). The combined influences of these variables suggested groomed
trails presented a novel habitat in which coyotes would
experience minimal threat from other predators and low
resistance to winter travel.
Additionally, groomed trails received a relatively high
ranking for prey encounters (rank = 6 out of 18,
mean = 29.0 prey encounters/km). Although other habitats
ranked higher, one should consider that, because of low
snow depth and high level of supportiveness in this habitat,
coyotes could cover more distance in a shorter time,
expending less energy and encountering more prey due to
123
38
J Ethol (2014) 32:29–41
Fig. 2 Relationship between canopy cover and significant biological
covariates of interest (Table 5): relationships between canopy cover
and snow depth (top left panel), snow penetration (top right panel),
snowshoe hare track encounter rate (bottom left panel), red squirrel
track encounter rate (bottom center panel), and ungulate track
encounter rate (bottom right panel)
temporal constraints than compared to other habitats. It is
also possible, based on the distance ratio (which showed
coyotes were taking more direct travel routes when entering and exiting this habitat), that they may be using
groomed trails to primarily travel, possibly to access other
habitats with desirable prey or locate kills. Overall,
groomed trails had the most desirable traits for any habitat
encountered, suggesting that it could be the best habitat for
minimizing energy expenditures and maximizing returns.
While snow depth was noticeably low in WB3 habitats
and could provide the primary explanation for why coyotes
used this habitat more than available (i.e., ease of travel), it
should also be mentioned that diet analyses showed a high
presence of whitebark pine seeds in the diet of coyotes
during certain months (Dowd and Gese 2012). Because of
stand structure and maturity of these trees (their ability to
produce cones), combined with low snow depths (making
access to seed caches more available), access to whitebark
pine seeds would be advantageous for coyotes. Whitebark
pine seeds are an important food source for several bird and
mammal species including black bears, grizzly bears, and
red squirrels (Mattson and Reinhart 1997). If coyotes use
this resource with minimal energy expenditure and high
energetic gain, the observed use versus availability analyses could reflect a preference for older whitebark pine
habitats. In addition, lodgepole (especially trees similar in
structure and age to LP2) were also found in or adjacent to
WB3 habitats. While hunting and traveling in LP2 was
likely easier than in any of the other lodgepole habitats, and
could explain coyote use of this habitat, proximity to
whitebark pine could enhance coyote selection of LP2 by
association when foraging on whitebark pine seeds.
Significantly less use of habitats LP1 and SF0 both
suggest there are characteristics making these habitats less
desirable for coyotes than other habitats. As suggested
above, hunting and traveling maybe have been inhibited in
LP1 due to stand structure, as it is categorized as a very
dense, even-aged stand. As for SF0, it is possible that a
high predator encounter rate (mean wolf encounters = 1.01/km) could account for the difference in use
versus availability (Table 2). In North America, interference competition with wolves can be an important factor
influencing the distribution and abundance of coyotes
(Thurber et al. 1992; Peterson 1995; Berger and Gese
2007).
While open woodlands were ranked first in habitat use
by coyotes, they still used open woodlands less than was
available on the landscape (use = 25.6 %, n = 337;
availability = 38.2 %). Several factors likely influenced
this avoidance. High levels of snowmobile traffic and
human presence (Dorrance et al. 1975; Richens and Lavigne 1978; Eckstein et al. 1979; Hamr 1988; Gander and
Ingold 1997) occur in these open meadows, as this open
landscape provides off-trail snowmobiling. Low prey track
encounters (rank = 16 out of 18) was likely due to the
deep snow (rank = 16 out of 18) limiting the availability
123
J Ethol (2014) 32:29–41
of small mammals (Wells and Bekoff 1982; Halpin and
Bissonette 1988; Gese et al. 1996b) and hindering coyote
movement in deep snow (Crete and Lariviere 2003). In
addition, the high incidence of wolf tracks (rank = 18 out
of 18, thus least desirable) in the open woodlands increased
the likelihood of encountering wolves, and the coyotes
thereby possibly avoided this habitat to reduce the risk of
intraguild predation (Thurber et al. 1992; Berger and Gese
2007).
Coyote travel distance within habitats
As we hypothesized, distances traveled within habitats
were related to snow supportiveness, suggesting the cost of
locomotion influenced distance traveled within more
energetically expensive habitats. Coyotes traveled further
and straighter within habitats with more supportive snow,
while coyote travel paths were more convoluted in habitats
with less supportive snow. Canopy cover also had this
effect on coyote travel distance, indicating coyotes traveled
more distance (i.e., had a more convoluted travel path) in
habitats with denser canopy cover. Essentially, this indicated that coyotes were using forested habitats (with less
compacted snow) to hunt and non-forested habitats to travel. The effect of snow depth on distance traveled (coyotes
traveled farther on more supportive snow when snow
depths increased) supported this assumption, suggesting
coyotes changed their behaviors to minimize energy
expenditure in the presence of deeper snow.
Canopy cover and habitat use
The influence of canopy cover on habitat use was perhaps
one of the most important variables for predicting prey use
by coyotes. Canopy cover provides refuge for prey species
and can increase survival (Litvaitis et al. 1985). While prey
availability can be higher in forested habitats (Richer et al.
2002), coyotes are known to have the best hunting success
in open habitats (Gese et al. 1996b). However, deep snow
and compacted surfaces can limit prey availability and
hinder hunting success in open habitats during the winter
(Halpin and Bissonette 1988) forcing coyotes to adopt
other strategies for acquiring prey (Gese et al. 1996a). In
this regard, forested habitats could be advantageous to
coyotes in our study area, as dense canopy cover yields
lower snow accumulation on the forest ground, possibly
making prey detection and acquisition more attainable in
forested habitats during the winter than other habitats
containing deep snow and compacted surfaces. Although it
has been suggested that coyotes may be poorly adapted for
hunting in forested habitats (Richer et al. 2002), if use of
forested habitats is restricted to winter use and coyotes
have access to open habitats during the spring, summer,
39
and fall months, use of forested habitats during the winter
may be beneficial. Gese et al. (1996b) reported capture
success rates of prey by coyotes to be higher in forested
habitats, even though lower capture rates, lower detection
rates, and fewer predation attempt rates were demonstrated
by coyotes hunting in forested habitats. However, these
data were obtained from an area where snow compaction
and persistent human disturbance was not an issue during
prey acquisition in open habitats (Gese et al. 1996b).
Coyotes in our study area demonstrated versatility to
deep-snow conditions based on documented habitat use and
behaviors associated with that use. During our study,
coyotes appeared to be abundant and effectively used deepsnow habitats despite a light, non-supportive snow column.
Coyotes have been shown to use compacted trails to negate
the impacts of deep snow (Murray and Boutin 1991;
Murray and Larivière 2002; Bunnell et al. 2006). In our
study area, open woodland and groomed trails both had
open canopies. Similar to our interpretation of hunting in
dense canopies, we found coyotes used both open woodlands and groomed trails primarily for travel due their
consistency in traveling straight-line projections. Similar to
Thibault and Ouellet (2005), as snow supportiveness
increased, coyote use of open canopy habitats increased,
likely to minimize energy expenditure by traveling on more
supportive surfaces. The deeper the snow, the more we
observed coyotes using open habitats. This was likely due
increased expenditures in dense habitats where snow was
less compacted. As hypothesized, habitat use as a function
of canopy cover resulted in preferential selection of open
canopy covers for travel due to supportive snow characteristics, while dense canopy covers appeared to provide
the most profitable strategy for winter foraging.
Our results have management implications for agencies
charged with lynx recovery. Whether coyote use of these
deep-snow habitats will impact other species in the ecosystem is unknown, but recovery of Canada lynx into these
high-elevation areas could be jeopardized by increased
competition with coyotes (Bunnell et al. 2006). Use of
groomed trails within deep-snow environments may enable
coyotes more access to a broader variety and expanse of
habitat patches. When considering increased access to
forested habitats, forests provide some of the best habitat
for snowshoe hares (Litvaitis et al. 1985), and snowshoe
hares are a major food item found in lynx and coyote diets
throughout North America (Crete et al. 2001). Limiting the
expanse of groomed trail systems may minimize coyote
encroachment into these deep-snow environments.
Acknowledgments Funding and logistical support provided by: US
Department of Agriculture, Wildlife Services, National Wildlife
Research Center, Logan, Utah; US Forest Service, Bridger-Teton
National Forest, Jackson, Wyoming; and US Forest Service, Intermountain Region, Ogden, Utah; and Endeavor Wildlife Research
123
40
Foundation, Jackson, Wyoming. We thank D. Chi and K. Johnson of
the US Forest Service for their support, J. Bissonette and J. Squires
for review of the manuscript, and P. Dowd, S. Dempsey, M. Greenblatt, S. Hegg, M. Holmes, M. Linnell, S. McKay, and G. WorleyHood for field assistance.
References
Bartel RA, Knowlton FF (2004) Functional feeding responses of
coyotes, Canis latrans, to fluctuating prey abundances in the
Curlew Valley, Utah, 1977–1993. Can J Zool 83:569–578
Berger KM, Gese EM (2007) Does interference competition with
wolves limit the distribution and abundance of coyotes? J Anim
Ecol 76:1075–1085
Bernier PY, Swanson RH (1992) The influence of opening size on
snow evaporation in the forests of Alberta foothills. Can J For
Res 23:239–244
Bolker BM (2008) Ecological models and data in R. Princeton
University Press, Princeton
Bunnell KD, Flinders JT, Wolfe ML (2006) Potential impacts of
coyotes and snowmobiles on lynx conservation in the intermountain west. Wildl Soc Bull 34:828–838
Collett D (2002) Modelling binary data (second edition). Chapman
Hall/CRC, New York
Crete M, Lariviere S (2003) Estimating the costs of locomotion in
snow for coyotes. Can J Zool 81:1808–1814
Crete M, Ouellet JP, Tremblay JP, Arsenault R (2001) Suitability of
forested landscapes for coyotes in northeastern North America
and its implications for coexistence with other carnivores.
Ecoscience 8:2346–2354
Despain DG (1990) Yellowstone vegetation: consequences of environment and history in a natural setting. Roberts Rinehart,
Boulder
Dorrance MJ, Savage PJ, Huff DE (1975) Effects of snowmobiles on
whitetailed deer. J Wildl Manag 39:563–569
Dowd JLB, Gese EM (2012) Seasonal variation of coyote diet in
northwestern Wyoming: implications for dietary overlap with
Canada lynx? Northwest Sci 86:289–299
Eckstein RG, O’Brien TF, Rongstad OJ, Bollinger JG (1979)
Snowmobile effects on movements of white-tailed deer: a
case-study. Environ Conserv 6:45–51
Gander H, Ingold P (1997) Reactions of male alpine chamois
Rupicapra r. rupicapra to hikers, joggers, and mountain bikers.
Biol Conserv 79:107–109
Gese EM, Grothe S (1995) Analysis of coyote predation on deer and
elk during winter in Yellowstone National Park. Am Midl Nat
133:36–43
Gese EM, Rongstad OJ, Mytton WR (1987) Manual and net-gun
capture of coyotes from helicopters. Wildl Soc Bull 15:444–445
Gese EM, Ruff RL, Crabtree RL (1996a) Foraging ecology of
coyotes: the influence of extrinsic factors on a dominance
hierarchy. Can J Zool 74:769–783
Gese EM, Ruff RL, Crabtree RL (1996b) Intrinsic and extrinsic
factors influencing coyote predation of small mammals in
Yellowstone National Park. Can J Zool 74:784–797
Graybill FA, Iyer HK (1994) Regression analysis: concepts and
applications. Duxbury, Belmont
Halpin MA, Bissonette JA (1988) Influence of snow depth on prey
availability and habitat use by red fox. Can J Zool 66:587–592
Hamr J (1988) Disturbance behavior of chamois in an alpine tourist
area of Austria. Mt Res Dev 8:65–73
Koehler GM, Aubry MG (1994) Lynx. In: Ruggiero LF, Aubry KB,
Buskirik SW, Lyon LJ, Zielinski WJ (eds) The scientific basis
for conserving forest carnivores: American marten, fisher, lynx,
123
J Ethol (2014) 32:29–41
and wolverine in the western United States. US Forest Service,
Gen Tech Rep RM-254, Washington DC, pp 74–98
Kolbe JA, Squires JR, Pletscher DH, Ruggiero LF (2007) The effect
of snowmobile trails on coyote movements within lynx home
ranges. J Wildl Manag 71:1409–1418
Kuznetsova A, Brockhoff PB, Christensen RHB (2013) lmerTest:
tests for random and fixed effects for linear mixed effect models
(lmer objects of lme4 package). R-Version:1.1-0. http://cran.rproject.org/web/packages/lmerTest/index.html
Lesage L, Crete M, Huot J, Quellet JP (2001) Evidence for a trade-off
between growth and body reserves in northern white-tailed deer.
Oecologia 126:30–41
Lewis L, Wenger CR (1998) Idaho’s Canada lynx: pieces of the
puzzle. Idaho Bureau of Land Management, Tech Bull 98–11,
Boise
Litvaitis JA (1992) Niche relations between coyotes and sympatric
Carnivora. In: Boer AH (ed) Ecology and management of the
eastern coyote. Wildlife Research Unit, University of New
Brunswick, Fredericton, pp 73–86
Litvaitis JA, Sherburne JA, Bissonette JA (1985) Influence of
understory characteristics on snowshoe hare habitat use and
density. J Wildl Manag 49:866–873
Mattson DJ, Reinhart DP (1997) Excavation of red squirrel middens
by grizzly bears in the whitebark pine zone. J Appl Ecol
34:926–940
Murray DL, Boutin S (1991) The influence of snow on lynx and
coyote movements: does morphology affect behavior? Oecologia
88:463–469
Murray CD, Buttle JM (2003) Impacts of clear cut harvesting on snow
accumulation and melt in a northern hardwood forest. J Hydrol
271:197–212
Murray DL, Larivière S (2002) Canid foot size and regional snow
condition: evidence for selection against high foot load? Can J
Zool 256:289–299
Murray DL, Boutin S, O’Donoghue M, Nams VO (1995) Hunting
behavior of a sympatric felid and canid in relation to vegetative
cover. Anim Behav 50:1203–1210
Natural Resources Conservation Service (2008) SNOTEL data network. http://www.wcc.nrcs.usda.gov/snotel/snotel.pl?sitenum=
822&state=wy. Accessed Dec 2008
Neter J, Kutner MH, Nachtsheim CJ, Wasserma W (1996) Applied
linear statistical models. Irwin, Homewood
Patterson BR, Benjamin LK, Messier F (1998) Prey switching and
feeding habits of eastern coyotes in relation to snowshoe hare
and white-tailed deer densities. Can J Zool 76:1885–1897
Peterson RO (1995) Wolves as interspecific competitors in canid
ecology. In: Carbyn LN, Fritts SH, Seip DR (eds) Ecology and
conservation of wolves in a changing world. Circumpolar,
Edmonton, pp 315–324
Poulle M-L, Crete M, Huot J-P (1995) Seasonal variation in body
mass and composition of eastern coyotes. Can J Zool
73:1625–1633
R Development Core Team (2010) R. version 2.10.1 in R: a language
and environment for statistical computing. R Foundation for
Statistical Computing, Vienna
Richens VB, Lavigne GR (1978) Response of white-tailed deer to
snowmobiles and snowmobile trails in Maine. Can Field Nat
92:334–344
Richer MC, Crête M, Ouellet JP, Rivest LP, Huot J (2002) The low
performance of forest versus rural coyotes in northeastern North
America: inequality between presence and availability of prey.
Ecoscience 9:44–54
Shield J (1972) Acclimation and energy metabolism of the dingo,
Canis dingo, and the coyote, Canis latrans. J Zool 168:483–501
Talbot J, Plamondon AP, Lévesque D, Aubé D, Prévos M, Chazalmartin F, Gnocchini M (2006) Relating snow dynamics and
J Ethol (2014) 32:29–41
balsam fir stand characteristics, Montmorency Forest, Quebec.
Hydrol Process 20:1187–1199
Thibault I, Ouellet JP (2005) Hunting behavior of eastern coyotes in
relation to vegetation cover, snow conditions, and hare distribution. Ecoscience 12:466–475
Thurber JM, Peterson RO, Woolington JD, Vucetich JA (1992)
Coyote coexistence with wolves on the Kenai Peninsula, Alaska.
Can J Zool 70:2494–2498
US Fish and Wildlife Service (2006) Number of wolves and packs in
Wyoming, summer 2006, outside Yellowstone National Park.
Wyoming Game and Fish Department, Cheyenne
US Forest Service (1989) Bridger-Teton National Forest: final
environmental impact statement. US Department of Agriculture,
Forest Service, Washington, DC
41
Warton DI, Hui FKC (2011) The arcsine is asinine: the analysis of
proportions in ecology. Ecology 92:3–10
Wells MC, Bekoff M (1982) Predation by wild coyotes: behavioral
and ecological analyses. J Mammal 63:118–127
Wyoming Department of State Parks and Cultural Services (2008)
Wyoming State Government annual report. State Library
Division Publications and Marketing, Cheyenne
Zub K, Szafranskal PA, Konarzewskil M, Redman P, Speakman JR
(2009) Trade-offs between activity and thermoregulation in a
small carnivore, the least weasel Mustela nivalis. Proc R Soc
Lond B 276:1921–1927
123